Fama French Factor Attribution: Separating Beta from Alpha

Key Takeaways

• The three-factor model adds SMB (small minus big) and HML (high minus low book-to-market) to the market factor, explaining most stock and bond return variation.
• In a worked example, 29.9 percent of 31 percent cumulative excess return came from systematic size and value factor loadings, leaving only a statistically insignificant alpha.
• A common investor mistake is treating strong absolute performance as skill without stripping out factor exposures available cheaply in passive ETFs.
• Factor attribution connects directly to fee conversations: if returns are explained by accessible risk premia, paying active management fees is difficult to justify.

What It Is

The original Fama-French three-factor model adds two equity factors to the market factor of the Capital Asset Pricing Model:

• SMB (Small Minus Big), a long-short portfolio of small-cap minus large-cap stocks.
• HML (High Minus Low), a long-short portfolio of high book-to-market (value) minus low book-to-market (growth) stocks.

The 1993 paper also included two bond factors (term and default) and showed that the full five-factor set explains most of the variation in stock and bond returns. Later work by Fama and French (2015) extended the equity model to five factors by adding profitability (RMW) and investment (CMA). Carhart added momentum (MOM) in 1997.

Factor attribution uses a time-series regression of the portfolio's excess return on these factors. The betas tell you which risks the portfolio loads on. The intercept, alpha, is what is left after the factors are accounted for.

The Intuition

Brinson attribution asks about weights and picks. Factor attribution asks about risks. A manager might look brilliant against a large-cap benchmark because they tilt to small-cap and value stocks, both of which have historically earned premia. Those premia are available to any investor who buys a cheap small-cap value ETF. Paying an active fee for them is paying for beta dressed up as alpha.

Factor attribution strips the systematic exposures out. What remains is either security-specific skill, exposure to an unmodeled factor, or luck. The exercise forces both the manager and the allocator to be honest about what the fee is buying.

How It Works

For a monthly window, the three-factor regression is:

Rp - Rf = alpha + b_mkt * (Rm - Rf) + b_smb * SMB + b_hml * HML + e

Where:

Rp - Rf = portfolio return in excess of the risk-free rate
Rm - Rf = market excess return (e.g. CRSP total market index minus T-bill)
SMB     = small-minus-big factor return
HML     = high-minus-low book-to-market factor return
alpha   = unexplained intercept
e       = residual (idiosyncratic)

Run the regression on 36 to 60 months of portfolio returns. The estimated betas show the portfolio's loadings. Multiplying each beta by the factor's realized return over the period gives the factor contribution to the portfolio's excess return. The alpha, if statistically significant, is the part the factors do not explain.

Factor return data for US equity (and many international markets) is published and updated monthly on the Kenneth French Data Library at Dartmouth. Commercial vendors (MSCI, Axioma, Barra, Bloomberg) offer proprietary factor models with different constructions and additional factors (quality, low volatility, liquidity).

Worked Example

A small-cap value fund's monthly excess returns are regressed on three years of Fama-French factors. The fitted output:

alpha   = 0.05% per month, t-stat 0.6   (not significant)
b_mkt   = 1.02
b_smb   = 0.85
b_hml   = 0.60

Over the 36 months, cumulative factor realizations were: market excess +22%, SMB +6%, HML +4%.

Factor contributions to cumulative excess return:

• Market: 1.02 x 22% = 22.4%
• Size: 0.85 x 6% = 5.1%
• Value: 0.60 x 4% = 2.4%

Explained excess return: 29.9%. If the fund's realized excess return was 31%, the residual (roughly 1.1%, plus compounding details) is alpha in the statistical sense, but the tiny t-stat says the evidence for skill is weak. Most of what looked like a good absolute performance came from persistent small-cap and value loadings, both available passively. That finding reframes the fee discussion.

Common Mistakes

1. Running too short a window. Three-year windows on monthly data (36 observations) give noisy betas. Five-year windows are standard. Weekly or daily regressions can help on liquid portfolios but introduce non-synchronous trading issues.

2. Ignoring model specification. The three-factor model is a starting point. For portfolios with momentum or profitability tilts, adding MOM, RMW, and CMA is essential. An unmodeled factor shows up in alpha and overstates skill.

3. Survivorship bias in the factor returns. When using proprietary factor series, confirm that the underlying universe includes delisted and bankrupt names. Biased factor series flatter regression results.

4. Confusing in-sample fit with forward-looking edge. A high R-squared means the model explains past returns well. It does not mean the alpha will persist. Holding out the most recent 12 months and checking that alpha survives is a reasonable discipline.

5. Applying US factors to non-US portfolios. SMB and HML returns differ across regions. Using US factors to analyze a European or emerging market portfolio mislabels regional market risk as alpha. Use region-specific or global factor sets from the French library or a commercial provider.

Frequently Asked Questions

Q: What is Fama-French factor attribution in simple terms? It is a regression method that breaks a portfolio's returns into exposures to well-known systematic risk factors, market risk, small-cap premium, and value premium, plus a residual alpha. It answers whether a manager's returns came from taking known, accessible risks or from genuine stock-picking skill.

Q: How does Fama-French factor attribution affect investment decisions? If a manager's alpha disappears once you account for factor exposures, you are paying active fees for beta you could get from a cheap ETF. Factor attribution reframes fee negotiations by showing how much of the return is attributable to skill versus systematic premia that are available passively.

Q: What is a real-world example of Fama-French factor attribution? A small-cap value fund shows a cumulative 31 percent excess return over three years. Factor regression assigns 22.4 percent to the market, 5.1 percent to the size premium, and 2.4 percent to the value premium. The residual alpha is 1.1 percent with a t-statistic of 0.6, meaning the evidence for true skill is statistically weak.

Q: How can investors use Fama-French factor attribution when evaluating managers? Request factor attribution for any active manager claiming alpha. Check whether the regression window is at least 36 to 60 months and whether the model includes all relevant factors (momentum, profitability) for the strategy. A statistically significant alpha after full factor adjustment is meaningful; a large alpha on a sparse three-factor model may reflect an unmodeled exposure.

Q: How is Fama-French factor attribution different from Brinson attribution? Brinson attribution splits returns by sector weights and stock picks using the portfolio's own segmentation. Fama-French uses a statistical regression to identify systematic risk factor loadings. Brinson explains where within the portfolio the active return came from; factor attribution explains what kind of risk drove it.

Sources

1. Fama, E.F. and French, K.R. (1993). "Common risk factors in the returns on stocks and bonds." Journal of Financial Economics, 33(1), 3-56. https://www.sciencedirect.com/science/article/abs/pii/0304405X93900235
2. Fama, E.F. and French, K.R. (1993). Full text PDF. https://www.bauer.uh.edu/rsusmel/phd/Fama-French_JFE93.pdf
3. Kenneth R. French Data Library. Dartmouth College. https://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html
4. CFA Institute. "Industry Codes and Standards." https://rpc.cfainstitute.org/codes-and-standards

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.